How do approval-rating methodologies (Gallup, FiveThirtyEight, RealClearPolitics) affect comparisons between presidents?
Executive summary
Approval-rating series from Gallup, FiveThirtyEight (and its successor projects), and RealClearPolitics use different data collection, weighting and averaging rules — producing different point estimates and trends even from the same underlying polls. Gallup reports its own national telephone or mixed-mode tracking and historical series (daily or periodic depending on the era) [1]; FiveThirtyEight-style averages weight polls by methodology and historical accuracy and often prefer adult samples [2] [3]; RealClearPolitics publishes unweighted averages of available polls that mix adults, registered and likely‑voter samples [4] [2]. These methodological choices systematically affect cross‑presidential comparisons because earlier presidents are measured mainly by Gallup only, while modern presidents are represented in multiple averages [5] [1].
1. How the basic methods differ — phone tracking versus averaged pools
Gallup publishes a continuous, internally consistent job‑approval time series that historically used periodic multiday telephone polls and more recently weekly Gallup Daily tracking for certain presidencies, giving a single-method baseline across decades when Gallup was the only major source [1]. By contrast, FiveThirtyEight’s approach (and Silver Bulletin, a direct descendant) constructs daily averages that weight polls by pollster quality and methodology and explicitly combine adults, registered and likely‑voter surveys in defined ways [2]. RealClearPolitics typically reports an unweighted mean of the available polls, which mixes poll types without methodological weighting [4]. Those differences mean one series can show a bump or decline that another mutes or amplifies simply because of which polls are counted and how they’re combined [2] [4].
2. Sampling frame matters — adults vs. registered vs. likely voters
FiveThirtyEight and related analysis have documented that polls using registered or likely‑voter screens often produce different approval numbers than polls of all adults; in the case of Donald Trump, registered/likely‑voter polls tended to be more favorable than all‑adult polls, and automated/online polls also showed systematic differences [3]. RealClearPolitics’ mix of poll types will therefore reflect those underlying sampling-frame differences without correction [4]. Gallup’s long‑running telephone or mixed‑mode surveys aim for a consistent adult population over time, which helps when comparing presidents across eras — but only where Gallup was the sole source historically [1] [5].
3. Weighting and pollster quality change the picture
FiveThirtyEight-style averages weight more reliable pollsters more heavily and adjust for recency and sample design; Silver Bulletin reiterates this practice, stressing pollster ratings and methodology as core to its average [2]. RealClearPolitics’ unweighted mean gives equal influence to small, possibly lower‑quality polls and large national trackers, which can tilt averages on any given day [4]. That explains why final “average” approval numbers published by FiveThirtyEight descendants and RealClearPolitics sometimes differ even when they use many of the same polls [2] [4].
4. Historical comparisons are constrained by available data
For presidents before Bill Clinton, Gallup is often the only continuous source referenced in compilations, so cross‑presidential comparisons can reflect Gallup’s methods rather than a true apples‑to‑apples mix of modern averaging approaches [5] [1]. Wikipedia’s presidential approval page notes that for earlier presidents only Gallup results exist in those datasets, which means modern multi‑poll averages cannot be retroactively applied to those presidencies [5]. Analysts comparing presidents must therefore decide whether to use a single pollster’s consistent historical series (Gallup) or to compare modern averaged series while accepting uneven historical coverage [1] [5].
5. Practical effects: different numbers, different narratives
The choice of series matters for headline claims: Gallup’s current point readings (for example, a 41% approval snapshot for Trump in October 2025) come from its own polling and sample design [6]; RealClearPolitics’ rolling mean can report a somewhat different figure for the same period because it averages many contemporaneous polls [4] [7]. FiveThirtyEight/Silver Bulletin averages, by down‑weighting lower‑quality or nonrepresentative polls, can again show a distinct trend line [2]. Journalists and historians therefore can reach different conclusions about “best” and “worst” presidents depending on which series — and which methodological assumptions — they prioritize [1] [2] [4].
6. How to compare presidents responsibly — recommended checklist
Use a consistent source across the comparison (e.g., Gallup alone for long‑run historical comparisons) or disclose mixing rules if using an aggregate [1] [5]. Note the sampling frame (adults vs. registered vs. likely voters) because it systematically shifts results [3]. When using averages, report whether polls were weighted by pollster quality and explain the weight scheme [2]. Finally, present multiple series side‑by‑side where possible so readers can see methodological sensitivity rather than being persuaded by a single number [2] [4].
Limitations: available sources do not mention a single universally accepted “best” method; they document tradeoffs among Gallup’s consistency, FiveThirtyEight’s weighting for quality, and RealClearPolitics’ transparent but unweighted averaging [1] [2] [4].