Open data · CC BY 4.0 what each one claims, and what replicated no ads · no tracking
Bias Atlascognitive biases

Common method bias

also known as — Common method variance · Common source bias · Mono-method bias · Percept-percept inflation
Common method bias is the claim that two constructs measured by the same instrument, usually one self-report survey, correlate partly because of the shared method.
Does it replicate? Mixed The measurement has been repeated and agrees; the interpretation has been repeated and does not, with estimates of how much the bias distorts a correlation running from 18% of item variance to a 92% inflation of observed relationships.
VerdictMixed
First published2003
FieldBelief and probability
Sources19
Last checked2026-09-21
What it claims

What does Common method bias mean?

This is a property of a research design rather than a habit of a mind, which it shares with selection bias and survivorship bias and with little else in this collection. A respondent who answers questions about job satisfaction and about supervisor behaviour in one sitting brings something to both answers that belongs to neither construct: an acquiescent style, a wish to look consistent, a mood, the position of the items on the page.

Podsakoff, MacKenzie, Lee and Podsakoff define common method variance as "variance that is attributable to the measurement method rather than to the constructs the measures represent". Doty and Glick separate two things the literature runs together. Common method variance is how much of the variance in a measure the method accounts for. Common method bias is "the divergence between observed and true relationships among constructs", which is the quantity a reader of a results table actually cares about.

The two are not the same number and can move in opposite directions. Citing Cote and Buckley's 1988 work, Podsakoff and colleagues note that method effects inflate an observed relationship when the correlation between the methods is higher than the correlation between the measures with method effects removed, and deflate it when the correlation between the methods is lower. The response styles that supply much of this variance are the subject of the response bias entry, which is not repeated here.

Verdict

Repeats disagree, or the effect holds in some conditions and not others. The other 248.

The field
72 of the 192 entries in belief and probability share this verdict.
Does it replicate?

Has Common method bias been retested?

Mixed The measurement has been repeated and agrees; the interpretation has been repeated and does not, with estimates of how much the bias distorts a correlation running from 18% of item variance to a 92% inflation of observed relationships.

There is no single experiment here to re-run, so the closest available thing is re-analysis of the same multitrait-multimethod archive by different teams. On the size of the method component they converge. The 2012 review's Table 1 gives Cote and Buckley 26% method variance across 70 matrices, Williams, Cote and Buckley 25% across 11, Buckley and colleagues 22% across 61, Doty and Glick 32% across 28, and Lance and colleagues 18% across 18, a span of 18% to 32%.

On what that does to a research finding they split. Lance, Dawson, Birkelbach and Hoffman put method variance at 18%, lower than earlier reviews, and report that although common method variance does inflate observed relationships, "this effect is almost completely offset by the attenuating effect of measurement error". Doty and Glick, from the same kind of data, get a 26% bias they call cause for concern.

Podsakoff and colleagues, applying their own decomposition to Cote and Buckley's figures, calculate inflation of about 45%, and list 38% from Buckley and colleagues, 92% from Doty and Glick and 60% from Lance and colleagues. They concede the estimates overlap and are not independent, and that the spread runs from 38% to 92%. They also note that a single-method design implies a method correlation of 1.00, on which the same estimates would run from 94% to 270%.

Crampton and Wagner tested the pattern across the published record instead, analysing 42,934 correlations in 581 articles. They found general evidence of percept-percept inflation, and on a subsample of 11,710 correlations concluded it "has not had the broad, comprehensive effects envisioned by critics", so that "domain-specific investigations are required".

Read off Lance, C. E., Dawson, B., Birkelbach, D., & Hoffman, B. J. (2010). Method Effects, Measurement Error, and Substantive Conclusions. Organizational Research Methods, 13(3), 435-455.

Examples

What are some examples of Common method bias?

The engagement survey that predicts itselfillustration

One questionnaire asks staff to rate their manager, and twenty items later to rate their own commitment. Whatever a person brings to both answers, such as the mood they were in that morning, sits inside the correlation the study then reports.

The respondent working down the pageillustration

Someone filling in a long form agrees mildly with everything in order to finish it. Every scale on the form picks up that habit, so two scales can move together because of how the answers were given rather than because the things measured are related.

The correlation that changes when the source doesillustration

A team collects the outcome from records and keeps only the predictor on the survey. The relationship comes out a different size, which is the method effect becoming visible. It may come out larger, because a shared method can hide a relationship as readily as manufacture one.

Cases marked as illustrations describe nobody in particular. They are there to make the pattern recognisable, not to report that a particular thing happened.

The experiments

What experiments is Common method bias based on?

The design

The quantitative base is a small set of meta-analyses that refit confirmatory factor models to previously published multitrait-multimethod matrices. Cote and Buckley examined 70 published datasets and report in their abstract that "on average, traits account for less than 50% of the variance in construct measures". The figure everyone quotes from that paper, 26.3% of the variance in a typical research measure, is read here from Podsakoff and colleagues' 2003 account, which calls it variance "due to systematic sources of measurement error like common method biases" rather than to method alone.

The same account records that the share varied by field, lowest in marketing at 15.8% and highest in education at 30.5%, and by construct, 22.5% for typical job performance measures against 40.7% for attitude measures. Podsakoff, MacKenzie and Podsakoff summarise five such meta-analyses in their 2012 review and conclude that "18% to 32% of the total variance in the items used in these studies was due to method factors".

Doty and Glick assessed every multitrait-multimethod matrix published over 12 years in six social science journals and report that 46% of the variation in measures is attributable to the constructs, 32% to common method variance, and that this "results in a 26% bias in the observed relationships among constructs". Their own verdict on that figure is that it "is cause for concern but does not invalidate many research findings".

The largest single number in the literature comes from comparing sources rather than decomposing variance. Podsakoff and colleagues pooled meta-analyses that had tested rating source as a moderator and found the average corrected correlation between leader behaviours and outcomes was 0.456 from the same source and 0.191 from different sources. Their Table 2 defines its inflation percentage as the ratio of the two corrected correlations and gives 239% for that pairing, with six others at 212%, 148%, 208%, 133%, 304% and 184%.

The running text of the same review gives 213% and 147% for the second and third of those, so the table and the prose disagree on two figures. The authors add that some of the gap may be raters seeing different things rather than method bias. Item placement has been manipulated directly. Weijters, Geuens and Schillewaert modelled the correlation between all pairs of 76 items on data from 2,850 respondents, and as their result is reported in the 2012 review, two items measuring unrelated constructs correlated 0.04 when six items apart and 0.09 when adjacent.

The conditions are common. That review reports Bodner's finding that 76% of studies across six areas of psychology used a single measurement method. A 2025 study made the point without a matrix. Joel, Sakaluk, Kim, Khera, Qin and Stanton wrote ten relationship items with the key noun replaced by a plant or a mineral, such as "My partner and I have excellent taro", and called the scale Pseudo.

In a pilot of 331 people one factor accounted for 58% of the variance and all ten items loaded above .70. Among 477 people in long-term relationships, recruited as 495 and reduced by 18 who did not qualify, Pseudo correlated .43 to .58 with satisfaction, commitment, appreciation and trust, against .60 to .80 among those four measures themselves.

14 primary research5 commentary and critique

Origin

Who first described Common method bias, and when?

Year2003
WhoPhilip M. Podsakoff, Scott B. MacKenzie, Jeong-Yeon Lee and Nathan P. Podsakoff
WhereJournal of Applied Psychology

"Common Method Biases in Behavioral Research: A Critical Review of the Literature and Recommended Remedies", 88(5), 879 to 903. The paper did not coin the idea and says so, opening with the observation that "discussions of the potential impact of common method biases date back well over 40 years" and citing Campbell and Fiske's 1959 multitrait-multimethod matrix. What 2003 established was the checklist form in which reviewers now apply the concern. The numbers it rests on are older, from Cote and Buckley's 1987 meta-analysis of 70 construct-validation datasets.

The claim, and the replication this entry reads its figures from.

Where it runs out

When does Common method bias not apply?

No replication in the ordinary sense is available. Searches of Crossref, PubMed and the open web on 21 September 2026 for a registered report or a multi-site repetition of a common method bias finding returned re-analyses of shared archives, simulations and new demonstrations rather than repetitions; OpenAlex answered single records but rate-limited the search queries that day. Two recent studies are empirical rather than archival.

Joel and colleagues preregistered the dyadic half of their Pseudo work, and Rubenstein, Simon, Kammeyer-Mueller, Corwin, Morrison and Whiting report "an initial study and a constructive replication" in which a psychological separation task of at least 7.5 minutes attenuated correlations about as much as temporal separation or distinct sources did. Their full text was not obtained and only the PubMed abstract informs this entry.

The sceptical case is that the standard framing gets the mechanism wrong. Spector's paper argues that the popular position "suggesting CMV automatically affects variables measured with the same method is a distortion and oversimplification of the true state of affairs, reaching the status of urban legend", and proposes retiring the term in favour of "measurement bias that is the product of the interplay of constructs and methods by which they are assessed".

His 1987 study of affective and perceptual work measures found "little evidence for method variance as a biasing problem" and concluded that properly developed instruments of that type resist it, while adding that their validity cannot be assumed from that result. Only the abstracts of both papers were obtained, the 2006 one at the University of South Florida repository, which carries no file, and the 1987 one at OpenAlex.

Spector, Rosen, Richardson, Williams and Johnson later set out the positive version of the argument, that shared sources of method variance inflate a correlation while unshared sources attenuate it. The remedies are weaker than the diagnosis. Richardson, Simmering and Sturman ran 691,200 simulated datasets against three post hoc corrections and doing nothing. Where no common method variance existed, correction in most conditions produced less accurate estimates than no correction, and all three techniques falsely identified bias about half the time when the true correlation was above zero.

The exceptions matter: with an ideal marker the correlational marker beat no correction at a true correlation of zero, and the confirmatory factor analysis marker beat it above zero. Across all conditions the unmeasured latent method construct approach identified the presence or absence of common method variance correctly about 41% of the time, the correlational marker approach about 69%, and the confirmatory factor analysis marker 73%, rising to about 84% with ideal markers only.

Their summary is that they "cannot recommend any post hoc CMV technique as a means for correcting CMV's potential effects", though they do suggest the confirmatory factor analysis marker as a way to test for its presence. Fuller, Simmering, Atinc, Atinc and Babin note that many business researchers test for the problem and "almost no one reports problematic findings", and their simulation concludes that at typical reliabilities and effect sizes common method variance "does not represent a grave threat to the validity of research findings".

Podsakoff, Podsakoff, Williams, Huang and Yang answered in 2024 with a review titled "Common Method Bias: It's Bad, It's Complex, It's Widespread, and It's Not Easy to Fix". Unpaywall and Semantic Scholar both record it as hybrid open access under a CC BY licence at the publisher, and on 21 September 2026 the publisher's article page, its PDF, the Ingenta mirror and the SSRN copy all returned HTTP 403 pages carrying a Cloudflare managed bot challenge rather than a subscription notice. The article is open and was still not obtained, so only its Crossref abstract informs this entry and none of its figures are quoted here.

Often confused with

Which biases is Common method bias confused with?

Each of these is named in this entry's own prose as something it gets taken for. Where the two came out differently, that is worth knowing before you quote either.

Commonly misread as

How is Common method bias commonly misread?

The first error is treating a same-source design as proof that a result is inflated. Direction is not fixed. Podsakoff and colleagues' 2003 paper states that the evidence from Cote and Buckley's 70 studies "indicates that the observed relationship between a typical predictor and criterion variable is understated by approximately 26% because of common method biases", while their 2012 review calculates about 45% inflation from the same estimates.

Both wordings are on the page in the two papers, and a reader who has met only the phrase "inflated correlation" will find the pair surprising. Response styles cut the same two ways. In Baumgartner and Steenkamp's data, as the 2012 review reports it, a true positive correlation paired with negatively correlated response style components fell by 55%.

The second error is reading Harman's single-factor test as a control. Podsakoff and colleagues are blunt that the procedure "actually does nothing to statistically control for (or partial out) method effects", that it is at best an insensitive diagnostic, and that finding more than one factor "is not evidence that the measures are free of common method variance".

They do not recommend it, and it remains the most reported remedy. The third error is dismissing a moderation result as method artefact. Siemsen, Roth and Oliveira show that common method variance can inflate or deflate a bivariate linear relationship depending on how symmetrically it hits the measures, and that "quadratic and interaction effects cannot be artifacts of CMV", though both can be severely deflated by it.

An interaction found in single-source data is the one result the objection cannot touch. The fourth error is reading Spector as denying that method affects measurement. His argument is about sufficiency, that the method alone does not produce the bias and that constructs and methods interact, which is why his proposed replacement is more design work rather than less.

The fifth is reading Joel and colleagues' nonsense scale as a clean win. Pseudo predicted later satisfaction, commitment, appreciation and trust when the week 1 measure was controlled, but with the week 2 measure controlled it predicted satisfaction at p = .176 and commitment at p = .060.

Sources

Sources for Common method bias

  1. 1Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879-903. Full text read from a PDF mirror at scispace.com and checked against the authors and affiliations printed on it; Unpaywall records the article as closed at the publisher. The Cote and Buckley percentages quoted in this entry are this paper's report of them.primary
  2. 2Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012). Sources of Method Bias in Social Science Research and Recommendations on How to Control It. Annual Review of Psychology, 63(1), 539-569. Full text read from a course copy at the University of British Columbia. The figures attributed here to Bodner, Baumgartner and Steenkamp, Weijters and colleagues, and Williams, Cote and Buckley are this review's reports of them. Its Table 2 and its running text disagree on two of the same-source inflation percentages, and both are given above.primary
  3. 3Podsakoff, P. M., Podsakoff, N. P., Williams, L. J., Huang, C., & Yang, J. (2024). Common Method Bias: It's Bad, It's Complex, It's Widespread, and It's Not Easy to Fix. Annual Review of Organizational Psychology and Organizational Behavior, 11(1), 17-61. Abstract read at Crossref. Unpaywall and Semantic Scholar record the article as hybrid open access under CC BY at the publisher. On 21 September 2026 the publisher's article page, its PDF, the Ingenta mirror and the SSRN copy each returned HTTP 403 with a Cloudflare managed bot challenge, not a paywall, so the full text was not obtained.primary
  4. 4Spector, P. E. (2006). Method Variance in Organizational Research: Truth or Urban Legend? Organizational Research Methods, 9(2), 221-232. Abstract read at the University of South Florida repository, which holds a record and a link to the publisher but no file; full text not obtained.primary
  5. 5Spector, P. E. (1987). Method variance as an artifact in self-reported affect and perceptions at work: Myth or significant problem? Journal of Applied Psychology, 72(3), 438-443. Abstract read at OpenAlex; full text not obtained.primary
  6. 6Doty, D. H., & Glick, W. H. (1998). Common Methods Bias: Does Common Methods Variance Really Bias Results? Organizational Research Methods, 1(4), 374-406. Abstract read at Crossref and OpenAlex, which carry the 46%, 32% and 26% figures and the authors' verdict on them; full text not obtained.primary
  7. 7Crampton, S. M., & Wagner, J. A. (1994). Percept-percept inflation in microorganizational research: An investigation of prevalence and effect. Journal of Applied Psychology, 79(1), 67-76. Abstract read at OpenAlex, which carries the counts and both quotations; full text not obtained.primary
  8. 8Cote, J. A., & Buckley, M. R. (1987). Estimating Trait, Method, and Error Variance: Generalizing across 70 Construct Validation Studies. Journal of Marketing Research, 24(3), 315-318. Abstract read at Crossref and OpenAlex. The 26.3%, 15.8%, 30.5%, 22.5% and 40.7% figures quoted in this entry are Podsakoff and colleagues' 2003 report of this paper, not read from it.secondary
  9. 9Lance, C. E., Dawson, B., Birkelbach, D., & Hoffman, B. J. (2010). Method Effects, Measurement Error, and Substantive Conclusions. Organizational Research Methods, 13(3), 435-455. Abstract read at Crossref and OpenAlex, which carry the 18% figure and the offsetting claim; full text not obtained.primary
  10. 10Siemsen, E., Roth, A., & Oliveira, P. (2010). Common Method Bias in Regression Models With Linear, Quadratic, and Interaction Effects. Organizational Research Methods, 13(3), 456-476. Abstract read at Crossref and OpenAlex; full text not obtained.primary
  11. 11Richardson, H. A., Simmering, M. J., & Sturman, M. C. (2009). A Tale of Three Perspectives: Examining Post Hoc Statistical Techniques for Detection and Correction of Common Method Variance. Organizational Research Methods, 12(4), 762-800. Full text read from the copy in Cornell's eCommons repository.primary
  12. 12Fuller, C. M., Simmering, M. J., Atinc, G., Atinc, Y., & Babin, B. J. (2016). Common methods variance detection in business research. Journal of Business Research, 69(8), 3192-3198. Abstract read at Boise State University's research portal, which carries both quotations; full text not obtained.primary
  13. 13Spector, P. E., Rosen, C. C., Richardson, H. A., Williams, L. J., & Johnson, R. E. (2019). A New Perspective on Method Variance: A Measure-Centric Approach. Journal of Management, 45(3), 855-880. Abstract read at Crossref and OpenAlex; full text not obtained.primary
  14. 14Weijters, B., Geuens, M., & Schillewaert, N. (2009). The proximity effect: The role of inter-item distance on reverse-item bias. International Journal of Research in Marketing, 26(1), 2-12. Crossref confirms the record and carries no abstract; the paper was not opened, and its correlations are quoted here as Podsakoff and colleagues report them.secondary
  15. 15Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56(2), 81-105. Cited here only to date the antecedent Podsakoff and colleagues name. Crossref and PubMed confirm the record; the paper was not opened. Unpaywall and OpenAlex list a green open-access copy at doi:10.71955/duedata-2026-qqjerx, which resolves instead to an unrelated TU Dortmund dataset on eHealth literacy in German athletes, so that index entry is wrong.secondary
  16. 16Williams, L. J., Cote, J. A., & Buckley, M. R. (1989). Lack of method variance in self-reported affect and perceptions at work: Reality or artifact? Journal of Applied Psychology, 74(3), 462-468. Crossref confirms the record; OpenAlex carries only a French indexing note and the full text was not obtained. Its variance estimates are quoted above as Podsakoff and colleagues' 2012 review reports them.secondary
  17. 17Campbell, D. T., & Fiske, D. W. (1987). This Week's Citation Classic: Campbell D T & Fiske D W. Convergent and discriminant validation by the multitrait-multimethod matrix. Current Contents. A one-page retrospective the two authors signed on 11 February 1987. Read in full. It confirms the 1959 citation and says they set out "to demonstrate the pervasiveness of method variance".secondary
  18. 18Joel, S., Sakaluk, J. K., Kim, J. J., Khera, D., Qin, H. Y., & Stanton, S. C. E. (2025). Pseudo Effects: How Method Biases Can Produce Spurious Findings About Close Relationships. Psychological Science, 36(10), 780-795. Full text read twice from the CC BY copy at PubMed Central.primary
  19. 19Rubenstein, A. L., Simon, L. S., Kammeyer-Mueller, J. D., Corwin, E. S., Morrison, H. M., & Whiting, S. W. (2025). On the efficacy of psychological separation to address common method variance: Experimental evidence and a guiding research design framework. Journal of Applied Psychology, 110(10), 1297-1317. Abstract read at PubMed; Unpaywall records the article as closed and the full text was not obtained.primary

Every claim on this page was held against these sources on 2026-09-21. Nothing here is written from memory. Written and checked by Krishna Chagti; how entries are written and corrected. Found a mistake? Report it, with the sentence and a source.

About Common method bias

Is Common method bias real?

The measurement has been repeated and agrees; the interpretation has been repeated and does not, with estimates of how much the bias distorts a correlation running from 18% of item variance to a 92% inflation of observed relationships. The effect sizes from the original study and from the replication are printed side by side above, so the comparison is visible rather than asserted.