I85PREP
⌘ K
CA ANZ Data Analytics and Insights study materials
CA ANZ Data Analytics and Insights study resources

CA ANZ Data Analytics and Insights practice questions

One question at a time. Build your CA ANZ Data Analytics and Insights understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Data Fundamentals and Data Quality

Which data type is best suited to storing a customer's date of birth when only the date is needed for age calculations?

  1. Timestamp with time zone
  2. Free-text string
  3. Boolean
  4. Date Correct answer.

Explanation

A date type stores a calendar date without a time component, which is exactly what age calculations require and avoids spurious time-zone shifts. Storing it as text would prevent reliable date arithmetic and sorting.

Question 2Data Fundamentals and Data Quality

In a relational table, which constraint guarantees that each row can be uniquely identified?

  1. Default value
  2. Check constraint
  3. Foreign key
  4. Primary key Correct answer.

Explanation

A primary key enforces uniqueness and non-nullness on the chosen column or column set, so every row is uniquely identifiable. Other constraints govern relationships, allowed values, or fill behaviour rather than row identity.

Question 3Data Fundamentals and Data Quality

A dataset column records customer satisfaction on a scale of 1 to 5. Which measurement scale does this represent?

  1. Ordinal Correct answer.
  2. Interval
  3. Nominal
  4. Ratio

Explanation

Ordinal data has a meaningful order but the gaps between categories are not guaranteed to be equal. A 1-to-5 satisfaction rating captures ranking rather than measurable distance, so it is ordinal.

Question 4Data Fundamentals and Data Quality

Which characteristic describes structured data?

  1. It requires no schema and is stored as raw text
  2. It is captured only as audio and video files
  3. It can only be analysed by manual inspection
  4. It is stored in a predefined schema of rows and columns Correct answer.

Explanation

Structured data conforms to a predefined model, typically tables with defined columns and data types, which makes it easy to query with SQL. Unstructured forms such as text, audio, and video lack this fixed schema.

Question 5Data Fundamentals and Data Quality

What is the main purpose of a data dictionary?

  1. To schedule nightly extract jobs
  2. To encrypt sensitive columns at rest
  3. To store backup copies of the production database
  4. To document the meaning, type, and allowed values of each data element Correct answer.

Explanation

A data dictionary is metadata that describes each element's definition, data type, format, and permitted values so users interpret fields consistently. It supports governance and reduces ambiguity across teams.

Question 6Data Fundamentals and Data Quality

Which statement best defines data completeness?

  1. The degree to which data is available when needed
  2. The degree to which required values are present in a dataset Correct answer.
  3. The degree to which values match the real-world entity
  4. The degree to which values are unique across records

Explanation

Completeness measures whether all required fields and records are present, focusing on missing values. It is distinct from accuracy, availability, and uniqueness, which address other quality dimensions.

Question 7Data Fundamentals and Data Quality

A report shows a customer's age as 145 years. Which data quality dimension is most directly violated?

  1. Consistency
  2. Timeliness
  3. Completeness
  4. Accuracy Correct answer.

Explanation

An implausible age indicates the value does not reflect the real-world fact, which is an accuracy failure. Completeness, consistency, and timeliness concern presence, agreement across sources, and currency rather than correctness of the value.

Question 8Data Fundamentals and Data Quality

In data modelling, what does cardinality describe?

  1. The frequency of backup operations
  2. The number of relationships allowed between entity instances Correct answer.
  3. The number of distinct values in a column
  4. The physical storage size of a table

Explanation

Cardinality in modelling expresses how many instances of one entity can relate to instances of another, such as one-to-many. It constrains relationship rules rather than describing value counts or storage.

Question 9Data Fundamentals and Data Quality

A retail chain needs to combine sales transactions from 40 stores into a single reporting table with consistent column names and types. Which process does this describe?

  1. Data archiving
  2. Data integration Correct answer.
  3. Data profiling
  4. Data masking

Explanation

Data integration consolidates data from multiple sources into a unified, consistent structure for reporting. Profiling assesses quality, masking obscures sensitive values, and archiving moves inactive data aside.

Question 10Data Fundamentals and Data Quality

A table has a foreign key referencing a customer table. A row is inserted with a customer identifier that does not exist in the customer table. What happens if referential integrity is enforced?

  1. The insert succeeds and a new customer row is created automatically
  2. The insert succeeds and the foreign key is dropped
  3. The insert succeeds and the value is set to null
  4. The insert is rejected Correct answer.

Explanation

Referential integrity requires every foreign key value to match an existing parent key, so an unmatched value causes the insert to fail. Automatic creation or nulling would require triggers or cascade rules that are not implied by the constraint itself.

Question 11Data Fundamentals and Data Quality

During profiling, an analyst finds that 12 percent of postcode fields contain fewer than four digits. Which quality dimension is being assessed?

  1. Timeliness
  2. Consistency
  3. Validity Correct answer.
  4. Uniqueness

Explanation

Validity checks whether values conform to a defined format or rule, such as a four-digit postcode. Uniqueness, consistency, and timeliness address duplicates, cross-source agreement, and currency respectively.

Question 12Data Fundamentals and Data Quality

Which activity best describes data cleansing?

  1. Standardising formats and correcting or removing erroneous values Correct answer.
  2. Increasing the frequency of data loads
  3. Defining access roles for analysts
  4. Encrypting data before transmission

Explanation

Data cleansing detects and remediates errors such as inconsistent formats, duplicates, and invalid values so the data meets quality rules. Security and load scheduling are separate operational concerns.

Question 13Data Fundamentals and Data Quality

An analyst joins a transactions table to a customer table on customer identifier. Which join returns only transactions that have a matching customer?

  1. Cross join
  2. Full outer join
  3. Inner join Correct answer.
  4. Left outer join

Explanation

An inner join keeps only rows where the join key matches in both tables, so unmatched transactions are excluded. Outer joins retain unmatched rows, and a cross join produces every combination.

Question 14Data Fundamentals and Data Quality

What distinguishes a data lake from a traditional data warehouse?

  1. A lake requires all data to be transformed into a star schema before loading
  2. A lake prohibits storage of unstructured files
  3. A lake can only store numeric data
  4. A lake typically stores raw data in native formats with schema applied on read Correct answer.

Explanation

A data lake commonly lands raw data in native formats and applies structure at query time, whereas a warehouse typically loads transformed, schema-defined data. This makes lakes suited to varied and unstructured sources.

Question 15Data Fundamentals and Data Quality

A dashboard displays yesterday's inventory levels, but the business requires near real-time stock positions. Which quality dimension is most affected?

  1. Completeness
  2. Validity
  3. Timeliness Correct answer.
  4. Accuracy

Explanation

Timeliness concerns whether data is current enough for the decision it supports. A one-day lag fails a near real-time requirement even if the values themselves are accurate and complete.

Question 16Data Fundamentals and Data Quality

Two systems store the same customer's address, and the values conflict. Which data quality dimension is primarily at risk?

  1. Consistency Correct answer.
  2. Validity
  3. Precision
  4. Completeness

Explanation

Consistency requires that the same fact agree across sources and records. Conflicting addresses for one customer show a consistency problem, even if each value is well formed and present.

Question 17Data Fundamentals and Data Quality

Which technique helps detect duplicate customer records that differ slightly in spelling and spacing?

  1. Exact string equality on the primary key
  2. Fuzzy matching on name and address fields Correct answer.
  3. Aggregating records by country only
  4. Sorting records by creation date

Explanation

Fuzzy matching compares fields using similarity measures so near-identical names and addresses can be flagged as potential duplicates. Exact comparison misses minor variations, and sorting or coarse grouping does not detect similarity.

Question 18Data Fundamentals and Data Quality

An analyst needs to summarise total sales by region and month from a large transaction table. Which approach is most appropriate?

  1. Filter out all rows with a null region
  2. Aggregate sales with grouping on region and month Correct answer.
  3. Return every transaction row for manual review
  4. Join the table to itself repeatedly

Explanation

Grouping by region and month with an aggregate function produces the required summary efficiently. Returning raw rows or self-joining adds effort without producing the summary, and filtering nulls alone does not aggregate.

Question 19Data Fundamentals and Data Quality

A column intended to store a two-letter country code contains values such as 'AUS', 'United Kingdom', and 'nz'. Which issue is most evident?

  1. Missing values
  2. Inconsistent coding standards Correct answer.
  3. Excessive precision
  4. Outdated records

Explanation

The column mixes three-letter codes, full names, and lowercase abbreviations, showing no single coding standard is enforced. This is a validity and consistency issue rather than a missing or outdated data problem.

Question 20Data Fundamentals and Data Quality

Which practice best supports data lineage tracking?

  1. Relying on analysts' memory of past changes
  2. Storing only the final report in a shared drive
  3. Recording source, transformation steps, and destination for each dataset Correct answer.
  4. Deleting intermediate staging tables after each load

Explanation

Lineage documents where data originated, how it was transformed, and where it landed, enabling impact analysis and audit. Removing intermediates or relying on memory breaks the traceable chain.

Question 21Data Fundamentals and Data Quality

A dataset of 1,000 records has 40 records with a missing email address. What is the completeness rate for the email field?

  1. 60 percent
  2. 96 percent Correct answer.
  3. 40 percent
  4. 4 percent

Explanation

Completeness is the proportion of records with the required value present: 960 of 1,000 records have an email, giving 96 percent. The 40 missing records represent a 4 percent gap.

Question 22Data Fundamentals and Data Quality

A team normalises a sales table so that customer details are stored once in a separate customer table. Which benefit does this primarily provide?

  1. Reduced redundancy and fewer update anomalies Correct answer.
  2. Automatic encryption of customer data
  3. Faster execution of every possible query
  4. Guaranteed elimination of all data errors

Explanation

Normalisation removes repeated data by storing each fact once, which reduces redundancy and the risk of inconsistent updates. It does not eliminate all errors, universally speed up queries, or provide encryption.

Question 23Data Fundamentals and Data Quality

An analyst finds that a numeric field has a mean far above its median and a small number of extremely large values. Which action is most appropriate before modelling?

  1. Delete the median value from the dataset
  2. Convert the field to a boolean flag
  3. Assume the mean is representative and proceed
  4. Investigate the extreme values and consider transformation or robust statistics Correct answer.

Explanation

A mean well above the median signals a right-skewed distribution driven by outliers, so the analyst should verify those values and consider a log transform or median-based measures. Deleting the median or ignoring the skew would distort results.

Question 24Data Fundamentals and Data Quality

A quality rule states that an order's ship date must be on or after its order date. Which type of rule is this?

  1. Referential integrity rule
  2. Cross-field consistency rule Correct answer.
  3. Uniqueness rule
  4. Format validity rule

Explanation

The rule compares two fields within the same record to ensure they agree logically, making it a cross-field consistency check. Uniqueness, referential integrity, and format rules each address different concerns.

Question 25Data Fundamentals and Data Quality

A source system sends records with a load timestamp in local time without a time zone, and the warehouse stores timestamps in UTC. What is the most significant risk?

  1. Ambiguous timestamps that shift by daylight-saving changes Correct answer.
  2. Permanent loss of referential integrity
  3. Automatic duplication of every row
  4. Reduced column precision for numeric fields

Explanation

Local times without a zone identifier become ambiguous around daylight-saving transitions, so conversion to UTC can be off by an hour. This affects event ordering and reporting accuracy rather than keys, duplication, or numeric precision.

Question 26Data Fundamentals and Data Quality

Which statement best describes the purpose of a data dictionary?

  1. It encrypts sensitive fields before they reach the reporting layer
  2. It schedules the overnight jobs that refresh reporting tables
  3. It stores a full copy of every record loaded into the warehouse
  4. It documents the meaning, type and permitted values of each field in a dataset Correct answer.

Explanation

A data dictionary is reference metadata that records definitions, data types, formats and allowed values for fields, so users and systems interpret data consistently. It describes the data rather than holding it, moving it or protecting it, which are handled by storage, scheduling and security tools respectively.

Question 27Data Fundamentals and Data Quality

An analyst must summarise sales by region and month from a very large transaction table, and the business wants fast dashboard queries without repeatedly scanning the full table. Which approach is most appropriate?

  1. Delete transaction rows that are not needed for reporting
  2. Query the full transaction table live for every dashboard view
  3. Build a pre-aggregated summary table keyed by region and month Correct answer.
  4. Export the transaction table to a spreadsheet for pivoting

Explanation

Pre-aggregating the transactions into a summary table keyed by region and month reduces the data scanned at query time and gives predictable dashboard performance. Live full-table scans and manual exports do not scale, and deleting source rows destroys detail the business may later need.

Question 28Data Fundamentals and Data Quality

A warehouse loads a daily extract, but the source system occasionally sends the same order twice with different load timestamps. Which technique best prevents these duplicates from inflating revenue totals?

  1. Deduplicating on the business key and retaining the latest record Correct answer.
  2. Increasing the frequency of the daily extract
  3. Storing the load timestamp as a text field
  4. Adding a surrogate key to every incoming row

Explanation

Duplicates arise from repeated business events, so the fix is to identify records by their business key, such as order identifier, and keep only the most recent version. Surrogate keys, extract frequency and field data types do not resolve repeated business records and can even mask them.

Question 29Data Fundamentals and Data Quality

An analyst profiles a date-of-birth field and finds that 3 percent of values fall after the record creation date. Which data quality dimension is most directly affected?

  1. Timeliness
  2. Completeness
  3. Uniqueness
  4. Accuracy Correct answer.

Explanation

Accuracy concerns whether values correctly reflect the real-world fact they represent. A birth date later than the record creation date is impossible, so the value is wrong even though it is present, distinct and current.

Question 30Data Fundamentals and Data Quality

A team ingests events from three source systems into a lake. Each event must be traceable back to its originating system, file and load time. Which design choice best supports this requirement?

  1. Apply hashing to event identifiers before storage
  2. Attach source system, file name and ingestion timestamp as metadata on each record Correct answer.
  3. Compress each source file before landing it in the lake
  4. Store all events in a single untyped text file per day

Explanation

Lineage requires that each record carries metadata identifying where it came from and when it arrived, so downstream users can trace it to the source. Consolidating files, hashing identifiers and compressing files may help storage or privacy but do not record provenance.

Question 31Data Analytics Techniques and Tools

A retail analyst wants a single number that summarises the typical transaction value when a small number of very large corporate purchases skew the distribution. Which measure of central tendency is most appropriate?

  1. The mean
  2. The median Correct answer.
  3. The range
  4. The mode

Explanation

The median is the middle value once data are ordered and it is resistant to extreme outliers, so a few very large corporate purchases will not distort it. The mean uses every value in its calculation and is pulled upward by such outliers. The mode and range do not describe a typical central value in this sense.

Question 32Data Analytics Techniques and Tools

Which statement best describes the purpose of a data dictionary in an analytics project?

  1. It stores the raw records extracted from source systems
  2. It lists the users who are authorised to query the warehouse
  3. It documents each field's name, meaning, type and permitted values Correct answer.
  4. It records the schedule of automated report deliveries

Explanation

A data dictionary is metadata that defines the structure and semantics of each field, including its name, data type, description and any valid value constraints. This shared definition prevents analysts from misinterpreting fields. It does not hold the data itself, nor does it manage delivery schedules or user permissions.

Question 33Data Analytics Techniques and Tools

An analyst joins a customer table to an orders table on customer ID and finds that customers with multiple orders appear multiple times in the result. Which join type produced this outcome?

  1. An inner join Correct answer.
  2. A full outer join
  3. A left outer join
  4. A cross join

Explanation

An inner join returns every matching row combination, so a customer with three orders appears three times. This row multiplication is normal behaviour for a one-to-many relationship. The other join types differ in which unmatched rows they retain, but all matching-based joins replicate rows the same way; a cross join instead pairs every row with every row.

Question 34Data Analytics Techniques and Tools

A dataset of 200 customer ages is sorted ascending. The analyst wants the value below which 25 percent of ages fall. Which calculation is required?

  1. The standard deviation
  2. The interquartile range
  3. The third quartile
  4. The first quartile Correct answer.

Explanation

The first quartile, or 25th percentile, is the value below which roughly a quarter of observations lie. The third quartile marks the 75th percentile, and the interquartile range is the distance between the two. Standard deviation measures dispersion around the mean rather than a positional cut-off.

Question 35Data Analytics Techniques and Tools

A marketing team wants to test whether average spend differs between customers exposed to two different campaign designs. Which technique is most appropriate?

  1. A moving average
  2. An independent-samples t-test Correct answer.
  3. A chi-square test of independence
  4. Simple linear regression

Explanation

An independent-samples t-test compares the means of a continuous variable across two independent groups, which matches comparing average spend between two campaign groups. A chi-square test suits categorical association, regression models relationships between variables rather than group differences, and a moving average smooths a time series.

Question 36Data Analytics Techniques and Tools

Which practice best supports reproducibility when an analyst builds a data preparation workflow?

  1. Emailing the final output to colleagues for checking
  2. Manually editing the spreadsheet each reporting cycle
  3. Relying on memory to repeat the same steps
  4. Saving the transformation steps as version-controlled code Correct answer.

Explanation

Reproducibility requires that the same inputs and steps reliably produce the same outputs, which is achieved by capturing transformations as version-controlled code. Manual edits, memory-based repetition and ad hoc distribution cannot be audited or rerun consistently. Version control also records who changed what and when.

Question 37Data Analytics Techniques and Tools

An analyst needs to combine monthly sales files that all share the same column structure into one dataset. Which operation is designed for this?

  1. Appending the files vertically Correct answer.
  2. Joining the files on a key
  3. Pivoting the files to long format
  4. Filtering the files by region

Explanation

When files share identical columns, stacking their rows is a vertical append, sometimes called a union, which produces one longer table. Joining combines columns from tables matched on a key, pivoting reshapes columns into rows, and filtering removes rows. Only the append preserves all records in a single consistent structure.

Question 38Data Analytics Techniques and Tools

A column of postcodes is stored as text rather than a number. Which outcome is most likely if this is not corrected before analysis?

  1. Storage space is reduced compared with numeric types
  2. Leading zeros are preserved but arithmetic comparisons fail Correct answer.
  3. Aggregate functions automatically convert the values
  4. Postcodes sort numerically in the correct geographic order

Explanation

Text storage preserves leading zeros, which is often desirable for postcodes, but numeric operations and range comparisons will not behave as intended. Sorting becomes lexicographic rather than numeric, and aggregates such as averages are meaningless on an identifier. Data type choice should reflect how the field will be used.

Question 39Data Analytics Techniques and Tools

Which chart type is best suited to showing the distribution of a single continuous variable such as customer wait times?

  1. A pie chart
  2. A scatter plot
  3. A stacked bar chart
  4. A histogram Correct answer.

Explanation

A histogram groups a continuous variable into bins and displays frequencies, revealing shape, spread and skew. Pie charts show parts of a whole for categories, stacked bars compare category compositions, and scatter plots show relationships between two variables. Only the histogram directly profiles one continuous distribution.

Question 40Data Analytics Techniques and Tools

A model predicting loan default achieves 98 percent accuracy on a dataset where only 2 percent of loans default. Why is accuracy misleading here?

  1. A model predicting no default for every loan would score 98 percent Correct answer.
  2. Accuracy always overstates performance on balanced datasets
  3. Accuracy cannot be computed for binary outcomes
  4. Accuracy ignores the number of observations in the test set

Explanation

With a heavily imbalanced class distribution, a trivial classifier that always predicts the majority class attains accuracy equal to that class's prevalence. This makes accuracy a poor guide to whether the model identifies the rare but important default cases. Metrics such as recall, precision or the area under the ROC curve are more informative.

Question 41Data Analytics Techniques and Tools

Which activity in the analytics lifecycle involves confirming that a dataset meets agreed completeness and validity rules before it is used for reporting?

  1. Feature engineering
  2. Model deployment
  3. Data quality assessment Correct answer.
  4. Exploratory data analysis

Explanation

Data quality assessment checks data against defined rules for completeness, validity, consistency and accuracy before use. Exploratory analysis investigates patterns, feature engineering creates model inputs, and deployment releases models into production. Quality checks typically precede these downstream activities.

Question 42Data Analytics Techniques and Tools

An analyst writes a query to return customers whose total order value exceeds 1000. Which clause structure correctly filters on an aggregated total?

  1. WHERE SUM(value) > 1000 GROUP BY customer
  2. ORDER BY SUM(value) > 1000
  3. SELECT customer WHERE value > 1000 GROUP BY customer
  4. GROUP BY customer HAVING SUM(value) > 1000 Correct answer.

Explanation

Aggregate conditions must be applied after grouping, which is the role of the HAVING clause. A WHERE clause filters individual rows before aggregation and cannot reference aggregate functions in standard SQL. The correct pattern groups by customer and then filters the summed value.

Question 43Data Analytics Techniques and Tools

A time series of daily sales shows a repeating pattern every seven days. What is this pattern called?

  1. Random noise
  2. Autocorrelation of residuals
  3. Seasonality Correct answer.
  4. Trend

Explanation

A regular pattern that repeats over a fixed period, such as a week, is seasonality. Trend refers to the longer-term direction of the series, while random noise is the irregular residual variation. Residual autocorrelation describes dependence between successive errors rather than a repeating calendar effect.

Question 44Data Analytics Techniques and Tools

An analyst wants to reduce a dataset with 40 correlated numeric variables to a smaller set of uncorrelated components before modelling. Which technique is designed for this?

  1. K-means clustering
  2. A chi-square test
  3. Principal component analysis Correct answer.
  4. Logistic regression

Explanation

Principal component analysis creates new uncorrelated variables, called components, that capture as much variance as possible in fewer dimensions. Logistic regression predicts a binary outcome, k-means groups similar observations, and a chi-square test assesses categorical association. Only PCA performs the described dimension reduction.

Question 45Data Analytics Techniques and Tools

Which statement correctly distinguishes structured from unstructured data?

  1. Structured data conforms to a defined schema, while unstructured data does not Correct answer.
  2. Unstructured data cannot be stored in a database
  3. Structured data is always numeric, while unstructured data is always text
  4. Structured data is always smaller in volume than unstructured data

Explanation

The defining difference is schema: structured data fits predefined fields and types, whereas unstructured data such as free text, images or audio lacks that organisation. Unstructured data can be stored in databases or data lakes, and volume is not the distinguishing criterion. Structured data may include text fields.

Question 46Data Analytics Techniques and Tools

A dataset contains a numeric field with values clustered near zero and a few extremely large values. Which transformation is commonly applied to reduce this skew before modelling?

  1. A logarithmic transformation Correct answer.
  2. One-hot encoding
  3. Binning into equal-width intervals
  4. Standardising to a mean of zero

Explanation

A logarithmic transformation compresses the scale of large values and pulls in a long right tail, making the distribution more symmetric. Standardisation rescales but does not remove skew, one-hot encoding applies to categories, and equal-width binning discards information rather than correcting the shape.

Question 47Data Analytics Techniques and Tools

An analyst notices that a customer record has a birth year of 1823 and a status field containing the value 'unknown' in a field restricted to active or closed. Which data quality dimensions are affected?

  1. Accuracy of aggregation
  2. Uniqueness and consistency
  3. Validity and domain integrity Correct answer.
  4. Completeness and timeliness

Explanation

The implausible birth year fails a validity rule, and the status value lies outside the permitted domain. Both are violations of defined constraints on acceptable values. Completeness concerns missing data, uniqueness concerns duplicates, and consistency concerns agreement across sources.

Question 48Data Analytics Techniques and Tools

A dashboard shows monthly revenue by product line. Which design choice best supports rapid comparison of product lines within a month?

  1. A table of raw daily transactions
  2. A 3D pie chart with many slices
  3. A dual-axis line chart with unrelated scales
  4. A horizontal bar chart ordered by revenue Correct answer.

Explanation

An ordered horizontal bar chart makes relative magnitudes easy to compare because length is a highly accurate visual encoding and ordering aids scanning. 3D pie charts distort perception, dual axes with unrelated scales invite false comparisons, and raw transaction tables require the viewer to aggregate mentally.

Question 49Data Analytics Techniques and Tools

What is the primary purpose of holding out a validation or test set when building a predictive model?

  1. To increase the number of features available
  2. To speed up model training
  3. To remove outliers from the training data
  4. To estimate performance on data the model has not seen Correct answer.

Explanation

A held-out set simulates new, unseen data and provides an unbiased estimate of generalisation performance. Training on all data would let the model memorise patterns and overstate its real-world accuracy. Holding out data does not add features, speed training or clean outliers.

Question 50Data Analytics Techniques and Tools

An analyst calculates a correlation coefficient of 0.85 between advertising spend and sales. Which conclusion is justified?

  1. Increased advertising causes higher sales
  2. There is a strong positive linear association between the two variables Correct answer.
  3. The relationship is statistically significant at the 1 percent level
  4. Sales will rise by exactly 0.85 units per unit of spend

Explanation

A correlation of 0.85 indicates a strong positive linear association, but correlation alone does not establish causation or the size of any effect. Significance depends on sample size and a formal test, which the coefficient alone does not provide. The slope of a regression line would be needed to quantify the change in sales.