Straightlining and speeders in surveys

Straightlining is when a respondent gives the same answer to every item in a grid or a series of rating questions, regardless of what each item says, for example "4" all the way down a matrix. Together with speeders, people who finish far faster than anyone could read the questions, it is one of the main signs of low-effort answers that researchers check for and remove before analysis.

Both behaviors are forms of what Jon Krosnick called satisficing (Applied Cognitive Psychology, 1991): giving an answer that is good enough to move on rather than the most accurate one. Krosnick describes satisficing as more likely when the task is hard, motivation is low and the survey drags on.

Why it matters

Straightlined answers look like real data. They pull averages toward the middle or the top of the scale, inflate correlations between items (everything moves together), and can make a scale look more reliable than it is. Speeders add random noise. A few percent of such responses can change which driver looks most important or whether a difference between groups is significant.

How to detect straightlining

• Zero variance in a grid. Compute the standard deviation of each person's answers across the items of one matrix question. A value of 0 means identical answers.
• Longest string. Count the longest run of identical consecutive answers. Meade and Craig (Psychological Methods, 2012) list it among the standard indices for careless responding.
• Reversed items. Include one item worded in the opposite direction, such as "The app is hard to use" next to "The app is easy to use". Agreeing with both is a contradiction that an attentive respondent would not produce.

How to detect speeders

Compare each response's completion time with the median for the whole sample. Decide a cutoff before fieldwork, for example less than 40% of the median, and apply it the same way to everyone. Time per page is more precise than total time when your survey has branches of different length.

Worked example

A customer survey has an 8-item grid on a 1–5 scale and a median completion time of 9 minutes. The team set two rules in advance: a response is a speeder below 3.6 minutes (40% of the median), and a straight-liner if all 8 grid answers are identical and it also agrees with both the normal and the reversed item. Out of 820 responses they remove 38 speeders, 24 straight-liners, 11 duplicate responses and 17 incomplete or nonsense ones. That leaves 730 valid responses, an 11% removal rate. The survey data cleaning calculator does this arithmetic and shows the removal rate.

Not every straight line is careless

Someone who really likes your product may rate every positive statement "5". Removing them would bias the results downward. That is why it is safer to combine several flags before removing anyone: straightlining, speeding, a failed attention check, gibberish in open answers. Meade and Craig also advise using more than one index. One practical rule is to remove a response only when it has two or more flags, and to report how many were removed and why.

How to prevent it

• Shorter grids. Split long matrices into several smaller ones, or into separate questions on mobile, where grids are harder to answer.
• Mix the direction of items so that an attentive person cannot agree with everything.
• Randomize item order so that fatigue does not always hit the same items.
• Shorter surveys. Cut questions you will not analyze; see survey fatigue.
• Attention checks, used sparingly and written politely, such as "Please select 'Disagree' for this row".
• Fair incentives and good sources. Low-quality panels and pay-per-complete schemes attract speeders and bots; see fraud detection.

Common mistakes

• Making up rules after seeing the results. Cleaning criteria chosen to get a preferred answer distort the study. Write them down before fieldwork.
• Removing on a single weak signal, such as identical answers in a 3-item grid.
• Using total time only in a survey with long optional branches.
• Not recording exclusions. Keep the reason for each removal so the report is transparent and the rule can be reused in the next wave.

Checking data quality in SurveyNinja

Each response in SurveyNinja shows how long it took, and from the Basic plan you can filter responses by completion time to find speeders. A response you reject can be excluded from the report. Random order of questions and answer options, a password, a captcha and one response per person are available on every plan, including Free. For the straightlining checks, export the answers to CSV, XLSX or SPSS on any plan and compute the grid variance in a spreadsheet or statistics package.