First look at the data

Introduction to Social Research Methodology

Author
Affiliation

Ben Stanley

Department of Social Sciences, SWPS University

Published

January 12, 2027

Scenario

The Meridian staff survey is back. As the junior research team advising the board of Meridian — the mid-sized Polish retail-and-services company whose turnover has risen from 14% to 27% in two years — you now hold 84 completed questionnaires in a single data file. The board meets on Friday and wants first findings: who is at risk of leaving, and what distinguishes them?

Work in pairs, with one laptop per pair running Excel or Google Sheets. Tasks 1–3 use the laptop; Task 4 is done on paper — put the laptop away for it.

Download the data: meridian-staff-survey.csv

Open the file in your spreadsheet (in Google Sheets: File → Import → Upload; in Excel it opens directly). Check before you start that you see 84 rows of data plus a header row, and nine columns.

Codebook

Column Meaning Values
respondent_id anonymous respondent code R001–R084
division where the respondent works Retail / Ecommerce / HQ
contract contract type Full-time / Part-time
tenure_months months employed at Meridian integer, 1–180
age_group age band 18-24 / 25-34 / 35-44 / 45+
engagement “I feel engaged in my work at Meridian” 1 (strongly disagree) – 5 (strongly agree)
pay_satisfaction satisfaction with pay 1 (very dissatisfied) – 5 (very satisfied)
schedule_satisfaction satisfaction with working schedule 1 (very dissatisfied) – 5 (very satisfied)
intends_to_leave “I intend to leave Meridian within the next 12 months” Yes / No

There are no missing values. Before analysing anything, ask of each variable: what level of measurement is this? — it decides which statistics you may use.

Task 1: One variable at a time (15 minutes)

Univariate description first — always.

  1. Frequency table for intends_to_leave. Produce a table showing the count and the percentage of respondents answering Yes and No. Use a pivot table (Insert → Pivot table; put intends_to_leave in Rows and again in Values, summarised by COUNT), or COUNTIF. Report percentages to one decimal place, with the base N.
  2. Frequency table for division. Same again: count and percentage for Retail, Ecommerce, and HQ. Note which divisions are large and which are small — you will need this in Task 3.
  3. Mean and median tenure. Compute the mean and the median of tenure_months (=AVERAGE(...) and =MEDIAN(...)).
  4. Write one sentence answering: which of the two — mean or median — is the more informative summary of how long Meridian staff have been with the company, and why? (Hint: compare the two numbers. What does the gap between them tell you about the shape of the distribution?)

Task 2: Two variables together (15 minutes)

Now the relational questions the board actually cares about.

  1. Crosstab: intention to leave by contract type. Build a pivot table with contract in Rows, intends_to_leave in Columns, and respondent_id in Values (summarised by COUNT). Then convert the counts to percentages within each contract type (in Google Sheets: show values as % of row; in Excel: Show Values As → % of Row Total — or compute them by hand from the counts).
    • Write down: what percentage of full-time staff intend to leave? What percentage of part-time staff?
    • One sentence: which group is the higher flight risk, and how big is the gap?
  2. Group means: engagement of leavers vs stayers. Using a pivot table (intends_to_leave in Rows, engagement in Values summarised by AVERAGE) or AVERAGEIF, compute the mean engagement score of those who intend to leave and those who intend to stay. Report both means to two decimal places, with the group sizes.
    • One sentence: how large is the gap, judged against the width of a 1–5 scale?

Task 3: Three findings for the board (10 minutes)

Numbers are not findings until someone can act on them. Write three findings from Tasks 1 and 2 in plain managerial language, one or two sentences each. Every finding must follow the genre model — claim + number + caveat:

Model: “A substantial minority of staff are at risk of leaving: 37% of survey respondents (31 of 84) say they intend to go within a year. Since respondents who have already disengaged may be under-represented among survey completers, the true figure could be higher.”

Requirements:

  • each finding contains at least one specific number with its base N;
  • each finding contains an honest caveat (sample size, who might be missing, association vs causation, subgroup size — whichever genuinely applies);
  • no jargon: the board should understand every word without a methods course;
  • at least one finding must come from Task 2 (a relationship, not just a level).

Task 4: The suspicious table (10 minutes — no computer)

Put the laptops away. A regional manager has sent the board his own analysis from a small pilot survey of 60 employees in two Kraków stores, conducted before the company-wide survey. His table and conclusion are reproduced exactly below.

Intention to leave by overtime pattern, Kraków pilot (raw counts):

Intends to leave: Yes Intends to leave: No Total
Works regular overtime 16 24 40
No regular overtime 4 16 20
Total 20 40 60

Beneath the table, the manager has written:

“Fully 80% of the employees who intend to leave (16 of 20) work regular overtime. Overtime is thus overwhelmingly the dominant factor driving staff towards the exit, and eliminating it should virtually eliminate the problem.”

On paper:

  1. Compute the row percentages — within each overtime group, what percentage intends to leave and what percentage intends to stay? (Four percentages; show your working.)
  2. Which way did the manager percentage the table to obtain his 80% figure — within the rows (overtime groups) or within the columns (leavers vs stayers)? What question does his percentage actually answer?
  3. Rewrite his conclusion correctly in one or two sentences: state the comparison the right-way percentages support, using your numbers from step 1, and add one honest caveat about what this table cannot show.

Before you leave

Hand in (one per pair, names on top): your Task 1 sentence, your Task 2 percentages and means, your three Task 3 findings, and your Task 4 working and corrected conclusion.