This is the best book on Indonesia's 1999 elections yet written. Dwight King aims to “get beyond the stress on personalities” and to “shift the focus more toward policies and institutions” (p. 3). With this, King brings the study of Indonesia's politics into the mainstream. His trove of data, methods, and conclusions will interest Indonesia specialists and comparativists alike.

The book can be divided roughly into two parts. Most chapters deal with the evolution of Indonesia's new electoral institutions. Chapter 3, for example, explores the development of the post-Suharto party and electoral laws, highlighting the high level of contestation that went into the creation of the new system, as well as the many complementary reforms that were required in order to create a new political playing field. Similarly, Chapter 4 offers a retrospective on the 1999 vote, including an analysis of the numerous problems that arose as the new institutions were first tested. Further chapters in this vein offer a case study of the electoral institutions in operation in one community in Java, analyze institutional changes adopted since the 1999 elections, and compare the path of Indonesia's electoral institutions to those in neighboring democratizers, Thailand and the Philippines.

The second part of the book focuses on using quantitative analysis to explain the outcome of the 1999 vote (Chapters 6 and 7). This is King's greatest contribution, and it is where his best data is mobilized. In Chapter 6, he takes on the common belief that there was a “broad continuity” in the vote results from 1955 (when Indonesia's last free elections were held) and 1999 (p. 124). He grapples mightily with this notion of continuity and sets out to discover if “voters with certain characteristics … supported certain parties in each election, revealing or articulating sociocultural divisions in the electorate” (p. 124). For example, abangan (nominal) Muslims were believed to prefer Megawati Sukarnoputri's Indonesian Democracy Party–Struggle (PDI-P) inasmuch as that same type of voter had preferred the Indonesian National Party (PNI) associated with her father or the Indonesian Communist Party (PKI) 44 years earlier.

Assuming that a district's sociocultural composition would be broadly similar in 1955 and 1999, as these types of characteristics change only slowly, King tested correlations of support for major parties from 1955 with the major parties of 1999 by district. It is important to note that this was not a sample of districts. The author used vote results for most of Indonesia's districts, making this one of the most comprehensive comparisons of Indonesian elections ever conducted.

King's findings strongly support much of the “conventional wisdom” on Indonesia's politics. For example, as hypothesized above, PDI-P did in fact draw upon areas of past PNI and PKI support. As a further illustration, the Golkar party was believed to have “greened,” or become more Islamic, over the previous 20 years. King's analysis supports this proposition by showing that Golkar's support was positively correlated with past support for Islamic parties in 1955.

On some points, though, King's analysis corrects the conventional wisdom. It could be assumed that with its basis in the bureaucracy and as a party from the secular-nationalist political stream, Golkar might find support in areas that had previously supported the bureaucrat-based, nationalist PNI. This turned out not to be the case. In fact, 1955 support for PNI was negatively correlated with 1999 support for Golkar. King's analysis of all the major parties proceeds similarly and offers both confirmations and surprises. He concludes this analysis by finding that the New Order regime of Suharto “failed” in its attempts at using “political engineering” to reduce the salience of Indonesia's sociocultural divisions in voting (p. 134). The cleavages of 1955 reappeared in 1999.

The author begins a new strand of quantitative analysis in Chapter 7. He uses regression to build models explaining each of the major parties' 1999 vote. He does this relatively successfully, his models explaining more than half the variation in the parties' votes.

The most significant critique that can be offered of Half-Hearted Reform has to do with the assumption that one can correlate votes by area in 1955 and 1999 and draw conclusions about the parties that are chosen by different types of voters on this basis. Sociocultural characteristics change only slowly, but they do change, as they have in Indonesia. The glaring example is in the decline of “nominal” Muslims over the past 40 years.

King often uses the term abangan and finds, in the final analysis, that the “basic cleavage in the Indonesian electorate between areas supporting nationalist and religiously inclusive parties (abangan) and areas supporting Islamic parties (santri) has not disappeared” (p. 130). Santri refers to pious, orthodox Muslims. This is striking because several studies in recent years have shown that the abangan themselves are dwindling. This is the finding of an unpublished study cited by King but not engaged on this issue (R. William Liddle and Saiful Mujani, “The Power of Leadership: Explaining Voting Behavior in the New Indonesian Democracy”). Like King, Liddle and Mujani hoped to understand the reasons behind voter choice in the 1999 elections. Their chosen means was via postelection opinion polling. They found that today about 80% of Indonesian Muslims would be described as santri. About 20% would thus be described as nominal Muslims, but only a small share of the total (a little more than 3%) fits the traditional characteristics of the abangan Muslim, that is, not avidly engaged in Muslim religious practice and simultaneously engaged in religious observances associated with Indonesia's pre-Islamic past. Liddle and Mujani's findings on the abangan need to be further studied, but the suggestion of changes in the population's sociocultural characteristics, as is implied by the disappearance of the abangan, needs to be addressed by King.

As found by Liddle and Mujani, santri do not necessarily vote only for santri/Islamic parties. Of respondents who were characterized as santri, Liddle and Mujani found that more than 25% voted for PDI-P, a secular-nationalist party. Likewise, their study showed that 53% of santri voters over-all chose parties in the non-santri tradition. Non-santri voters did seem to fit the model more easily. They went overwhelmingly for the non-santri parties (81%).

With his study, King has decisively demonstrated that a pattern of voting by area exists. But he missed the chance to engage Liddle and Mujani's absent abangan to consider what the sociocultural changes mean for his methodology and the conclusions he hopes to draw about voter groups. Voters in formerly abangan areas continue to vote for abangan parties, but we do not yet have an explanation as to why.