Take on me-asurement (take measurement on): Teaching measurement and coding with one-hit wonders

Abstract Measurement in the social sciences involves assigning values to particular empirical cases and plays a vital role in the research process, but it can be tricky to teach. Unless an undergraduate social sciences research methods course includes the collection of original data, many of the issues of measurement can seem abstract and arcane to students. To help illustrate how a social science researcher goes from a conceptualization to assigning values to cases, I developed a lecture that centers on the question “which artists are one-hit wonders?” I used the rules developed by Chris Molanphy to help students see how we can develop applicable methods for any possible case and make a measure that is both reliable and valid. The example also helps illustrate the difference between categorical and numerical variables by comparing the categorization of artists as one-hit wonders and the relative numerical value of the chart position of the artist’s most successful song.

Measurement is an often-underappreciated part of the research process.Teaching measurement to social science undergraduates can be even trickier, particularly if the students are not engaging in their own data collection (Adriaensen et al., 2015).Students can have difficulty grasping how to go from empirical observations of the world to numerical values of specific cases.If they do not grasp the basics of measuring social and political concepts, the entire enterprise of quantitative methods can be opaque.When it comes to trying to understand concepts, a student's natural inclination can be to just rely on the promise of "I know it when I see it" without grappling with the difficulties of developing a measurement scheme that is clear, reliable, and valid.Social science students can also struggle with the distinction between continuous and categorical variables.
To illustrate the process of measurement in the social sciences, particularly how a researcher develops a coding scheme and applies it to a set of cases, I developed an exercise in my undergraduate political science research methods class based on the question "what is a one-hit wonder?"Before the class, I asked students to email me the name of the first musical artist that comes to mind when they think of the phrase "one-hit wonder."I then used the rules created by Slate journalist Chris Molanphy for what counts as a one-hit wonder to illustrate the importance of having a clear set of rules when measuring a concept, the difference between categorical and continuous variables, and false positives and negatives (Molanphy, 2012).Importantly, Molanphy provides enough detail for the particular artists so the instructor can quickly glean the necessary information.Most of the examples that I present in this paper are taken directly from Molanphy's work.The characteristics that define a one-hit wonder and Molanphy's rules work quite well together because he has done most of the legwork in both his article and his podcast.He not only clearly articulates the rules, but he explains why these rules exist, based on how they classify a set of artists that many would consider to be either an archetype of a one-hit wonder or feel insulted if they were labeled as such.The article and podcast have so many examples of the edge cases of the rules that instructors can plug and play with whatever their musical tastes are.They can also choose to avoid particular songs that are terrible.

Teaching the "how to" of measurement in the social sciences
The introduction to research methods classes for political science or social science undergrads have a lot of ground to cover.Many, including mine, start with the basics of what social science research is, go through research design, and end up with some relatively introductory statistics.This makes it exceedingly difficult to give enough attention to all the topics.In my course, students have a semester-long research project, but there simply is not enough time for them to collect their own data, so we rely on existing datasets (usually the most recent American National Election Study).One implication of this choice is that students do not get any experience in turning empirical observations of the world into the coding of specific cases.They do not have to directly measure any concept but take the measure that others have created.
Many of the major political science textbooks do not emphasize this process, either.They provide examples of how scholars have chosen to measure complicated topics.Kellstedt and Whitten (2018) use the examples of how to measure democracy (Munck & Verkuilen, 2002) and political tolerance (Stouffer, 1955;Sullivan et al., 1979).Pollock and Edwards (2019) give several examples of measurement, particularly from common survey questions.Johnson et al. (2015) discuss Hall and Miler's work on legislators' intervention into air pollution regulations (Hall & Miler, 2008); Segal and Cover's measurement of Supreme Court Justices' policy attitudes (Segal & Cover, 1989); Bradley et al.'s cross-national measures of income (Bradley et al., 2003); and Wattenberg and Brians (1999); and Ansolabehere et al. (1994) measures of campaign exposure.
In every case, these examples illustrate the main points that the textbook authors are trying to make, yet the concepts can be relatively arcane to students, particularly if they have not taken many political science courses.More importantly, the examples (with the exception of some of Pollock and Edwards' discussion) do not ask the students to think about how they would go from an empirical observation of the world to assigning a specific numerical value to a particular case.
The lack of direct experience with assigning values can also limit the intuitiveness of the other aspects of measurement.It's essential for social science students to understand the distinction between continuous and categorical variables, for instance, in order to choose the appropriate statistics that appear in these textbooks.While all the texts make this point to the reader, a direct application of assigning scores to cases may make the concept easier to digest.Reliability and validity can also be relatively esoteric to students if they don't grasp the basic idea of assigning numerical values to cases.

Using the "one-hit wonder" to illustrate measurement
The approach I took to illustrate these points was to apply them to the idea of a one-hit wonder in popular music.For those unfamiliar with the phrase, it is generally applied to musical artists who had one very popular song and never achieved that level of success again.It has a little bit of derision to it, with artists being dismissed as nothing but a one-hit wonder.It also is a term that is relatively ubiquitous in America but surprisingly vague.Most American undergraduates can recognize the term one-hit wonder and can think of musical artists that they think fit the term but have not given much thought to systematically define the term and apply it to a set of possible cases.This makes it a useful example of the problem in measuring whether or not an artist actually fits this term.
The lecture has three parts as illustrated in Figure 1.First, I introduce the concepts of measurement and one-hit wonders.I go through the artists the students suggested before class and a curated list of other plausible artists that will illustrate the key points of the lecture.Second, I develop the measurement rules that will be applied to each of the artists.Finally, I present the completed spreadsheet coding every artist and presenting the highest chart performance of any of their songs.

Lecture part one: Motivating the topic
The lecture begins by introducing the action of measurement; how a researcher goes from an empirical observation of a case in the world to a numerical value in a spreadsheet.In this context, a case is a musical artist, and the coding is 1) are they a one-hit wonder?and 2) what is the highest performance of one of their singles on the Billboard Hot 100?Students are presented with a spreadsheet that has columns labeled with the artists' names, a column labeled one-hit wonder, and a column labeled top chart performance.The columns themselves are empty.It is important to remind the students of the goal here.We need to be able to identify any possible artist in the history of popular music as either a one-hit wonder or not.
Students are then presented with the artists that they suggested as one-hit wonders.For this, I include YouTube clips of the highest-performing single from that artist.As an aside, instructors probably want to actually listen to the songs and, if necessary, find a radio-friendly version of the song.My students found the "radio version" amusing, but a few chuckles in class are probably preferable to a complaint to an administrator for the language in the songs.
It is relatively easy to find the chart performance of any artist or song.Wikipedia's discography entry for an artist tends to list all of the songs that have placed on any chart in any country.It is important to make sure that you are consistent in which chart you use and understand the differences.In many, but not all, cases, the Wikipedia article will include a link to Billboard's page for the artist.These links generally have the stem https://www.billboard.com/artist/ARTIST_NAME, where "ARTIST NAME" is replaced with the artist's name (https://www.billboard.com/artist/prince/for instance) and ends with the name of the artist.This provides a drop-down menu of all the charts they appeared on and the user can choose "The Hot 100" for singles or "The Hot 200" if you need to see how their albums charted (see below).The charts are in alphabetical order (charts that start with "The" are listed under the letter "T").
My students generally did an excellent job of choosing artists that allow me to illustrate the main points of the lecture, and it was easy to supplement the list they generated with artists that would illustrate key points.I suggested to the students that they could also determine which artists are one-hit wonders by relying on the views of "experts."While some fields would never rely on this type of measurement approach, they are common in political science.One of the most common measures of how democratic a country is (an example used throughout the course) comes from Polity (Polity, 2011).Polity relies on expert judgments about countries and years for hundreds of countries going back more than 100 years.)Forthis lecture and categorizing an artist as a one-hit wonder, I used VH1's list of 100 Greatest One-Hit Wonders. 1 VH1 is an American network that has its roots in broadcasting music videos.As the network has evolved, it morphed into frequently producing shows on specific topics like "Behind the Music."It is a prominent source of documentation about popular music in America and something that students are familiar with.This list provided several useful examples of key points, particularly A-ha's "Take On Me," Los Del Rio's "Macarena," Vanilla Ice's "Ice Ice Baby," Men Without Hats' "Safety Dance," Young MC's "Bust a Move," and Marky Mark and the Funky Bunch's "Good Vibrations."Most of these songs are ubiquitous enough to be familiar to most undergraduates.At this point in the lecture, students are often stating their disagreement with some of the choices.I use this to illustrate the concept of the reliability of the measure.Different coders are coding artists differently.The lack of clear rules, I tell them, means that this type of subjective coding can result in the same artist being measured differently.This also includes a very brief discussion of intercoder reliability that focuses on the concept more than an actual statistic like Cronbach's alpha, but one could include that here.
From here, I populated the example spreadsheet with the names of the artists that were examples of potential one-hit wonders that came from either the students or the experts.As each artist was, based on the rules, definitively categorized as either a one-hit wonder or not, I went back to this spreadsheet and coded them as a "zero" if they were not and a "one" if they were.At the same time, we added the highest chart performance of their best song.To make sure I did this correctly, I did not complete this in class but rather created a second version of the spreadsheet with these entries filled in.I used the same process for each possible step in this lecture.
Should we trust these experts?VH1 is opaque about how they came up with this list and several of these artists (including Vanilla Ice and Marky Mark) had more than one-hit, even if the followups were, retrospectively, regrettable and forgettable.VH1 was motivated by entertainment and viewers, not the same standards we should rely on as social scientists.The list is also not reliable.VH1 does not make its process clear, and it is unlikely that another panel of experts on music or culture would come up with the same list.I illustrate this to students by comparing their list to the VH1 list, noting that the differences illustrate a possible difficulty in reliably measuring this concept unless we have specific rules that could be objectively applied to a musical artist.
At this point, I tried to unpack the phrase "one-hit wonder."Each of the words has meaning, but not necessarily an obvious one in this context.Starting at the end, "wonder" is simply a placeholder for "artist."Yes, the wonder part of it implies that there is something fantastical about the artist.But some of these artists that are definitely one-hit wonders should never be described as fantastical. 2The use of wonder is also somewhat sarcastic and pejorative.Most artists and their fans may not like having this term applied to them.I point this out in the lecture noting that we are going to take social science separately, we need to be honest about measurement.We may not like the normative implications of the measurement scheme or our research results, but it is important to be intellectually honest in the process.
The second word, "hit," can be controversial, but there is a relatively straightforward way to objectively define what counts as a hit.Since 1955, the Billboard Hot 100 chart has been dedicated to tracking the most popular songs in America.As music technology has shifted, so has the Hot 100, 3 but the list provides a definitive and reliable measure of what counts as a hit.Billboard does have a large and changing number of hits, but the Hot 100 is widely seen as the chart of the top songs in America in any particular week since the 1950s. 4As Molanphy (2023) notes, one of the virtues of this Billboard list is that it combines what he refers to as active and passive fandom.Active fandom here means conscious and active choices by fans to consume the content through physical purchases or digital downloads.Additionally, the list has also always included radio play, which Molanphy refers to as passive fandom.Listeners who hear songs this way do not make the choice to listen, but still hear the song in their day to day lives.
Additionally, appearing on the Hot 100 is not enough to qualify as a hit.Instead, being in the Top 40 (potentially being counted down by Casey Kasem on his long running show, American Top 40) is held as the definitive proof that a song is in fact a hit.The importance of being in the top 40 to be a hit I so ingrained in American music culture, that "Top 40" has become a genre and label for American radio stations.If a suggested artist did not chart one song on the Top 40 of the Hot 100, they cannot be a onehit wonder.If students suggested such an artist, I turned to the spreadsheet and filled in that artist's "one-hit wonder" column with a "0."In part lectures, this has included Rebecca Black and her song "Friday" (which did not reach the Top 40) and Semisonic and their song "Closing Time," which did not chart on the Hot 100 because it was never released as a single and, therefore, not eligible (see footnote 5).
Note that the use of Billboard also excludes charts from other countries.For a lot of acts from outside the U.S., this might be unfair.They will often have several hits in their home country even if only one breaks through in the U.S. A-ha, for instance, has had 20 songs that were in the top 10 in at least one country, but they are often considered one of the definitive one-hit wonders in the U.S. It also excludes artists who may have had multiple hits in other genres and on other lists such as Country, R&B/Hip-Hop, or Alternative.These other charts are important but are more niche than the Hot 100 list.This leaves the word "one" in one-hit wonder.Presumably, an artist would be a one-hit wonder if they have one and only one song chart in the top 40 of the Billboard Hot 100.This one seems really obvious, but it is surprisingly complicated.It was at this point that I returned to the spreadsheet and coded the artists that could be excluded as one-hit wonders because they had a second song reach the top 40 of the Billboard Hot 100.A key example of this is, again, A-ha.While they are generally remembered for "Take On Me," they had the number 20 hit "The Sun Always Shines on T.V." It is a good song, but most do not know it and it feels wrong for A-ha to not be a one-hit wonder.This very simple rule creates another surprising complication.There are several artists that had only one song that charted on the Hot 100 but do not fit the typical view of what is a one-hit wonder.In particular, some artists have had long and successful careers that focus more on albums than singles.
The three examples I used in the lecture were Jimi Hendrix, 5 The Grateful Dead, 6 and Lou Reed. 7While all of these artists have been inducted into the Rock & Roll Hall of Fame, each of them only had a single song break into the top 40 of the Hot 100 (Molanphy, 2012).I presented the students with a new spreadsheet that included these three artists and coded them as one-hit wonders.These artists are generally not what we think of as a one-hit wonder and their inclusion illustrates both the notion of face validity and false positives.By those technical rules, each of these artists is classified this way, but very few people would accept that.They are false positives.This means we have to go back to a spreadsheet that lists the artists but leaves the coding of one-hit wonder as blank.More generally, I add a brief discussion of the validity of the measure here.It was a concept introduced in a previous lecture, so this is more of a reminder than a deep discussion of the concept.

Lecture part two: Explaining the rules
If the obvious definition of "one Top 40 song" excludes cases that are pretty clearly one-hit wonders (false negatives) and includes artists that are absolutely not one-hit wonders (false positives), then we need a better set of rules.At this point, I introduced Molanphy's three rules.
An artist is a one-hit wonder if they had one Top 40 hit, and they didn't meet any of the following three conditions: (1) They did not have a second Top 10 hit single. 8  (2) They did not have a second Top 40 single more than six months after their first single.
(3) They did not have three or more Top 10 or platinum albums.
These rules are deeply embedded in Molanphy's knowledge of charts.As will be discussed below, it results in a set of cases that minimizes the false negatives and positives and maximizes the face validity of the measures.
Then I took each of the rules and applied them to key cases.Rule 1 actually means that Vanilla Ice and Marky Mark and the Funky Bunch are not actually one-hit wonders.Vanilla Ice's follow-up single "Play that Funky Music (White Boy)" reached number 4 on the Hot 100 and Marky Mark's "Wildside" reached number 10.
Rule 2 is the key for including several artists that people tend to think of as one-hit wonders that would have been otherwise excluded.A-ha's second hit was from the same album and peaked only four months after the success of "Take On Me."They are not the only example: Young MC, Rachel Platten, 9 and a long list of other artists had a second song that reached the Top 40 as a follow-up to their breakthrough hit.Particularly when radio play dominated the makeup of the Top 40, a new artist could often get enough airplay from a second song no matter how bad the song was (Molanphy, 2012).Molanphy created this six-month rule to discount this effect and include artists who only matter in pop music for their single most memorable song.
The third rule resolves some of the false positives noted above.It excludes The Grateful Dead and Jimi Hendrix.Both of these artists had several prominent albums.Molanphy notes that this rule does step outside the view of songs as songs, independent from their full albums, but this rule is necessary for the face validity of the measure.Another example of an artist that almost needs this rule is Radiohead."Creep" was the number 34 song for one week in 1993.That would have been their only hit except in 2008, Radiohead had fans remix the song "Nude" using the separated stem tracks.The sales of these stem tracks gave the song enough of a boost in sales to reach number 37 for its only week on the Hot 100 chart.But for that technicality, they would have been a one-hit wonder by the obvious definition, yet no Radiohead fan would accept this.Rule 3 means they would not have to.Radiohead has had six Top Three albums, and they are clearly not a one-hit wonder.
To end the discussion of specific artists, I return to Lou Reed.None of Molanphy's rules exclude Lou Reed from being classified as a one-hit wonder."Walk on the Wild Side" was the only song he ever had that was on the Hot 100, let alone in the Top 40, and he had only one Top 10 album (Sally Can't Dance).The great irony that I explained to students is that the song that made Lou Reed a one-hit wonder is the heart of "Wildside," the song that meant that Marky Mark is not a one-hit wonder.
I used that final nugget to illustrate a final point about measurement.Once we have rules, we mat refine them because of false negatives or positives, but when the final criteria is established, we consistently apply them.I told the students of my love of Lou Reed's music and that, personally, it is kind of painful to put that "1" in the one-hit wonder column of the spreadsheet where his name appears.But that is how we have to approach our data and social science more generally.We may not like the answers we get, but we need to be intellectually honest about the process.Lou Reed is a one-hit wonder whether I like it or not.

Lecture part three: Applying the measurement scheme
The final step in the lecture was to go back to the spreadsheet with the third column listing the highest chart performance of every artist's top single.See Table 1 for an example of this spreadsheet with a list of artists from students and the ones I added.Several of these artists will have had number one songs and, if you chose the artists correctly, there could be a reasonable range of values for that item.I ended the lecture by comparing the information that we had in both of these columns.The zeros and ones in the one-hit wonder column were simple illustrations of the concept.The numbers in the cells just indicated that there was something different between the zeros and ones, but there wasn't anything inherent in the numbers.In contrast, the highest chart performance column provided actual numbers.At this point, the instructor is left with a choice.You can tell the students that these numbers are cardinal and that they have a specific meaning.A song at number four is twice as high as a song at number eight.In terms of chart performance, the gap between the fifth and the tenth most popular songs is as large as the difference between the 30 th and the 35 th .This, however, is not strictly true.The chart performance is really more of an ordinal measurement.We do not know that the differences between any two numbers on the chart are equal.This is a measure of how popular a song is in any week.The gaps between the songs are masked by the simple ordinal ranking.This may, however, be too complicated of a topic for the point you wish to make.If you want an actual continuous measure, you could use the sales of the single, yet this creates comparability problems as you compare songs from the era when the 45 single dominated to the era where streaming is most prevalent.Honestly, this is the very end of the lecture, and these points may be too arcane and are likely to be lost on many students.

Conclusion
This was a fast-moving lecture, and I covered a lot of ground, but it illustrated several key points about measurement that are a challenge to articulate.Students reacted very positively to it.They found the inclusion of songs interesting and my occasional commentary about the artists amusing.They were engaged in the process and were sheepishly disappointed to learn that an artist they had suggested didn't meet the technical definition.
The rules that Molanphy provides are clear and, in most cases, easy to make a definitive decision upon.One could build a lecture with just these rules without the same depth of knowledge of chart history that Molanphy possesses.The episode of the Hit Parade podcast, however, provides so much information and so many anecdotes about the chart history of all the potential one-hit wonders that it can provide the content for most of the lecture.
There are likely to be many other equally interesting examples of measuring something that students can identify.The presence of objective rules based on one particular chart makes this a relatively easy example to work with.This example would probably not work for every instructor, as there is a little bit of trying (and maybe failing) to be a "hip" professor to pull it off.But if it fits an instructor's teaching style, it can be an effective technique for conveying what can otherwise be difficult concepts for social science students.
One limitation to this approach is that one-hit wonder, with the appropriate rules, ends up being a much more straightforward measurement problem than many of the latent constructs that are prevalent in the social sciences.Democracy, war, class, prejudice, and a host of other central variables in the social sciences are never directly observed and researchers have to rely on various indicators of these concepts.Adding this to this lecture works if the instructor has time at the end.The one-hit wonder may be something that students are familiar with and think they know but is a much easier concept to measure than several of the key concepts in the social sciences.
had a rule that the song had to be released as a single to be eligible.The music industry in this era made the choice to not release many top songs as singles to encourage buyers to purchase the entire album.The particular example that came up in my class was "Closing Time" by Semisonic, which was suggested by a student.That song was never released as a single for sale and never actually charted on the Billboard Hot 100.It is, however, included as an example of a onehit wonder by Molanphy (see https://twitter.com/cmo lanphy/status/1494821973653966848).

Hendrix's version of Bob Dylan's "All Along the
Watchtower" reached number 20 in 1969.His nexthighest charting song was "Crosstown Traffic" which only reached number 52. 6.The Grateful Dead had several songs that were on the Hot 100, but "Touch of Grey" was their only song to make it to the Top 40 when it reached number 9 in 1987.7. "Walk on the Wild Side" was not only Lou Reed's only Top 40 song, but it was also the only song he recorded that reached the Hot 100.This includes his work with The Velvet Underground, which released several singles but never charted.8.This rule exists because Top 10 songs are exceedingly rare for artists.Chart performance is an ordinal scale and being in the top 10 is, by most subjective senses, a clear hit.Molanphy does not do much to justify this rule, but it is consistent with the broader understanding of the relative popularity of hits.9. Political scientists might be more familiar with the version of "Fight Song" that was produced shortly before the 2016 election as an homage to Hillary Clinton (https://www.youtube.com/watch?v= YttscNOoAjA).

Figure 1 .
Figure 1.Structure of the onehit wonder as measurement lecture.